High-Resolution View Synthesis In Camera-Projector Systems Using Compressive Dual Imaging
Ivan Ralašić, Matea Đonlić, Damir Seršić · 2020
In this paper, a method for high-resolution view synthesis in a calibrated camera-projector system is proposed. Camera’s resolution is significantly lower than the projector’s resolution. Compressive sensing and sum-to-one transformation are used for an efficient light transport matrix estimation. The light transport matrix and Helmholtz reciprocity enable dual imaging, i.e. generating virtual images from the projector’s viewpoint and virtual relighting of the observed scene. Virtual fringe profilometry enables 3D reconstruction of the observed scene. Performing 3D reconstruction in the dual configuration of the camera-projector system results in a significantly denser point cloud. Reconstructed 3D points are back-projected onto the camera sensor using the camera’s projective matrix, resulting in a high-resolution image, synthesized from the camera’s viewpoint. The proposed method overcomes the spatial resolution limit of the imaging sensor, resulting in significantly increased spatial resolution of the camera’s images.